Broken Hills mine memory crash

[gterror]

SLOT09.zip
debug.log

Every time i enter the mine, the damage from the gas, which ticks, crashes the game to desktop. Using the latest rpu version with no alteration to sfall.

[burner1024]

reporting crashes

[gterror]

It seems that when i take the script file from the original killap RP 2.3.3 and copy it to scripts folder, the crash will be gone. I thought that its not needed to install the original version?

[burner1024]

I thought that its not needed to install the original version?

That is correct. It is likely a newly-introduced one.
I can reproduce the crash. For the time being, working around it with the old script should do fine, with no side effects.

[burner1024]

I modified the script to get some debug: broken2.int.zip. For some reason, party_member_list_critters returns null pointers. I don't think it's supposed to do that.

procedure CauseDamage begin
   variable who;
   ndebug("test0");
   foreach who in party_member_list_critters begin
      ndebug("who is " + who);
      ndebug("dude is " + dude_obj);
      ndebug("who name is " + obj_name(who));
      if not wearing_protective_gear(who)
         and critter_kill_type(who) != KILL_TYPE_robot_kills
      then begin
         ndebug("test1");
         apply_radon_damage(who)
         ndebug("test2");
      end
   end
end

broken2: test0

broken2: who is 37695152

broken2: dude is 37695152

broken2: who name is Marcus

broken2: test1

broken2: test3
Loading sound file monitor.ACM...succeeded.

broken2: test4

broken2: test2

broken2: who is 0

broken2: dude is 37695152
Script Error: scripts\broken2.int: op_obj_name: obj is NULL
broken2: who name is Marcus

OPCODE ERROR: critter_inven_obj2() - argument #1 is null.
 > Script: scripts\broken2.int, procedure <couldn't find proc>.Script Error: scripts\broken2.int: op_obj_pid: obj is NULL
OPCODE ERROR: critter_inven_obj2() - argument #1 is null.
 > Script: scripts\broken2.int, procedure <couldn't find proc>.Script Error: scripts\broken2.int: op_obj_pid: obj is NULL
OPCODE ERROR: critter_inven_obj2() - argument #1 is null.
 > Script: scripts\broken2.int, procedure <couldn't find proc>.Script Error: scripts\broken2.int: op_obj_pid: obj is NULL
OPCODE ERROR: critter_inven_obj2() - argument #1 is null.
 > Script: scripts\broken2.int, procedure <couldn't find proc>.Script Error: scripts\broken2.int: op_obj_pid: obj is NULL
OPCODE ERROR: critter_inven_obj2() - argument #1 is null.
 > Script: scripts\broken2.int, procedure <couldn't find proc>.Script Error: scripts\broken2.int: op_obj_pid: obj is NULL

@FakelsHub @NovaRain any idea why that happens?

[NovaRain]

The crash address 0x42D940 is in critter_kill_count_type_. If I remove the critter_kill_type(who) != KILL_TYPE_robot_kills check, the crash will be gone, but I also notice other party members don't receive any damage.

I tried to recreate the crash like this:

// The instruction at 0042d940 referenced memory at 00000064
// The memory could not be read from
foreach who in party_member_list_critters begin
   display_msg("Name is " + obj_name(who));
   display_msg("Kill type: " + critter_kill_type(who));
   critter_dmg(who, 2, DMG_normal_dam BWOR DMG_NOANIMATE BWOR DMG_BYPASS_ARMOR);
end

Removing the critter_dmg line: the name and kill type are displayed for everyone (I got Sulik, Vic, and Cassidy in party).
Removing the critter_kill_type line: only the player's name gets displayed and the player receives 2 point of damage.
Leaving only the critter_dmg line: only the player receives 2 point of damage.

EDIT: and this crash seems to happen on all reasonably available versions (tried old 3.6, 3.8.x, 4.x)

EDIT 2: another simple test:

key := [1, 2, 3, 4, 5];
foreach i in key begin
   display_msg("Number " + i);
   critter_dmg(dude_obj, i, DMG_normal_dam BWOR DMG_NOANIMATE BWOR DMG_BYPASS_ARMOR);
end

The loop works only for the first time; the rest of it are all zero. It seems there's some conflict between critter_dmg and looping through an array.

[burner1024]

OK, it seems that the script isn't at fault then. I opened an issue in sfall repo.

[burner1024]

should be fixed now, with latest sfall